INDEX

Explanations

names of professions and roles

oai_token-act-pair · gpt-3.5-turbo

titles and roles within the entertainment and media industries

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 bleed

-0.61

 manic

-0.58

 anyway

-0.58

 Shutterstock

-0.56

 worse

-0.56

hedon

-0.56

 Grind

-0.55

gra

-0.55

 Yamato

-0.54

 dice

-0.54

POSITIVE LOGITS

lishes

0.67

arthed

0.62

omics

0.61

 specializing

0.60

assadors

0.60

uds

0.59

 alumni

0.58

 chaired

0.57

hid

0.57

 toured

0.57

Activations Density 1.471%

names of professions and roles

titles and roles within the entertainment and media industries

No Comments

No Known Activations